A panel of five seated at the front of a room

Design session surfaces instructor strategy on AI learning assistance

“What do we do now that we have AI?”

Danielle McNamara, executive director of the Learning Experience Institute at Arizona State University asked the question at a design session, hosted by the university. For many instructors, leveraging AI in teaching has become foundational in creating a constructive learning environment.

According to the National Assessment of Education Progress’ Long—Term Trend test, student math and reading scores have substantially declined. ASU, in collaboration with its partners, is working to reverse this trend by turning to the instructor’s needs.

The design session — which focused on expanding AI strategy in the classroom — took place on ASU’s downtown campus, bringing together nearly 50 experts from across the country in collaboration with teams from ASU’s EdPlus, Provost Office and Enterprise Technology. 

Bethany Weigele, chief innovation officer for EdPlus, and Elizabeth Reilley, executive director of AI Acceleration spearheaded the day-long event. 

“One of the things that we should focus more on is helping students learn how to learn,” McNamara said during the opening panel discussion. “We often think that smart people learn math, but there’s plenty of evidence that math makes you smarter.”

“How do we support the instructor to get better content, more engaging content, more engaging activities, to think about what they actually want the student to learn when provided those opportunities?” asked Anne Jones, vice provost for undergraduate education at ASU.

“Even though they exist, we are often not taking advantage of the tools that would allow us to recognize patterns in what students are doing right, to then adapt our course, because we don’t even really have the time or know how to adapt our course on the fly,” Jones said. “Technology might lead to a fantastic answer to this.”

From assignment assistance to personalized, data-driven learning, experts across the nation weighed in on potential use cases for AI in STEM-related courses.

Chris Rasmussen, professor for Department of Mathematics and Statistics Center for Research in Mathematics and Science Education at San Diego State University emphasized the enhancements AI can bring to homework and exam preparation for students.

According to Rasmussen, homework assignments in mathematics typically present problems in a “blocked,” style, highlighting a specific problem-solving technique for one block, followed by a new block introducing a new technique.

“Then the exam comes, and you don’t know which technique to use, because they're sort of intermixed,” Rasmussen said. “So, the alternative to block design is called, ‘interleave.’”

Mixing problem-solving techniques on homework would be highly effective in preparing students for intermixed problems on exams, Rasmussen said. The benefits include higher performance and longer retention of the newly learned skills.

“I think AI could probably help instructors figure out how to interleave, because that's a hard task,” Rasmussen said. “Reshuffling all your homework problems so that they're interleaved, that's a huge burden, and I think AI could help instructors with that.”

Some instructors would even like to see the interleave method expanded across classrooms, creating a learning ecosystem that continues running regardless of which course the student is taking next.

Thinking of how AI can improve the overall learning experience of students trickled into another common dilemma: student to teacher ratios in classrooms.

In one class alone, Jones instructs large course sections. According to her, it’s impossible to get personable enough with each student to fill in learning gaps on her own, as she can with her small group of PhD students.

“How might I be able to observe what is happening with students, the level of interaction I get with my PhD students, even in a cohort of thousands?” Jones asked. “That is, in fact, the strength of machine learning: recognition of patterns, identification of attributes, of data that I can't see as a human.”

For many instructors across the nation, teaching STEM-related courses, particularly math courses, comes with the unique challenge of tackling the difficulty of the subject while trying to reach students.

“AI can help us think about broadening the way in which students have to demonstrate competency, not just doing multiple choice tests,” McNamara said. “Rethinking those ties and rethinking how to build in good conduct and good learning theory, and leverage AI for that. AI can help us.”

In a discussion following the event, McNamara underscored the importance of the design session bringing together faculty and staff together to iterate ideas for AI in the classroom: “Very often what would happen would be the person creating the technology and then various people would come together and create something, and what you create is not anything that instructors or students wanted or needed. And so, part of that learning engineering process is the ideation stage, bringing stakeholders to think about what can be done, what should be done, what you would do, and what you would use.”